whatmodel.sh / QwQ-32B

QwQ-32B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

QwQ-32B is an open-weights language model from Alibaba, released March 2025. Its published size is 32.76B parameters and a 131k-token context window. Licence: apache-2.0.

Catalogue identity notes

Comparison model from the Epoch AI internal-evaluation CSVs (Epoch release date 2025-03-05). Release checkpoint Qwen/QwQ-32B (March 2025); not QwQ-32B-Preview. 32,763,876,352 parameters. ollama qwq:32b = 32b-q4_K_M (same digest).

59.4±0.8#202 of 719 LLMsfrom 27 benchmark results across 13 sources

$ how it compares

The LLMs ranked either side of QwQ-32B, out of 828 with a score, then more from Alibaba.

#modelscorewhereRAMreleased
242Mistral Small 4 (119B-A6.5B) Mistral60±2api $0.26/M +local~40GB2026-03
243Gemini 2.0 Flash-Lite Preview (02-05) Google60±3comparison only2025-02
244QwQ-32B Alibaba59±1local~21GB2025-03
245OpenReasoning-Nemotron-32B NVIDIA59±2local~21GB2025-07
246DeepSeek V3 (Dec '24) DeepSeek59±1api $0.46/M +local~337GB2024-12
more from Alibaba
18Qwen3.8 Max (2.4T-A95B) Alibaba78±1api $3.00/M +local~380GB2026-08
34Qwen3.8 Max (0902) Alibaba76±1api $3.00/M2026-09
37Qwen3.8-Flash-Next Alibaba75±1local~59GB2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #222 of 300 * · 2.9% of its score · 2 tests

Beats o1 (2024-12-17), o3-mini, Llama 3.1 Nemotron Ultra 253B v1 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, LFM2.5 1.2B Thinking here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL V4 (2025-11-03 archive, models absent from the December archive)36.9% ✎▃··▅▃▃·▅·▃·█ beats 3/10hard · avg 34%▇ 1.5%at its level
BFCL v358.3% ✎▁▁▁▁▁▃▅█▅▄▁▁ beats 75/114medium · avg 52%▇ 1.4%blunt · at its level
coding▁▁▁▃▆██▇▆▄▃▁ #203 of 458 · 28% of its score · 7 tests

Beats Qwen3.5 4B, Qwen3-VL 32B Instruct, Falcon-H1R 7B here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 3B Reasoning 2512, DeepSeek-Coder-V2 Instruct (0724) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Aider polyglot (225 exercises, two attempts)20.9% ✎▂·▂▃▅▂▃█▆▄▂▂ beats 11/49very hard · avg 8.5%▇▇▇▇▇▇▇ 17%sharp · at its level
WebApp1K-Duo-React pass@1 (onekq-ai)61.6% ✎▁··▁·▁▁▁▆█▇█ beats 16/45hard · avg 32%▇ 3.4%at its level
WebApp1K-React pass@1 (onekq-ai)87.0% ✎▃▁▁█▁▄▅▃▄▇▁▄ beats 23/31medium · avg 44%▇ 2.2%sharp · noisy
CodeReviewQA: change localisation (hard), invariant accuracy63.8% ✎█▂▁▁▂▂▂▃▅▅▅▇ beats 42/47medium · avg 36%▇ 2.1%sharp · at its level · noisy
CodeReviewQA: automated code refinement, exact match rate32.9% ✎▁▁▁▁▁▁▂▂▃█▅▄ beats 30/47hard · avg 33%▇ 1.6%at its level
CodeReviewQA: solution identification (hard), invariant accuracy77.2% ✎▃··▁▂▂▃▂█▃▄▂ beats 41/47medium · avg 54%▇ 1.2%sharp · noisy
CodeReviewQA: change type recognition, invariant accuracy60.9% ✎▁▁▁······▁▃█ beats 10/47easy · avg 67%▇ 0.0%blunt · at its level · noisy
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #164 of 393 · 23% of its score · 5 tests

Beats MiniMax-M2, MiniCPM5 2B, Granite 4.2 30B here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, LFM2.5 1.2B Thinking, LFM2.5 8B-A1B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Arena-Hard v2.0 hard prompts (style control, Gemini-2.5 judge)43.5% ✎▂·▃▅▅·▃█▆▂▂▂ beats 11/22very hard · avg 12%▇▇▇▇ 10%sharp · at its level
LMArena (text)1335.8 ✎▁▁▁▂▃▃▃▆▆▇█▃ beats 174/379avg 1268▇▇▇ 6.4%blunt
AlpacaEval 2 length-controlled82.4% ✎▃▃·▁▁█▇▅·▁·▃ beats 22/22medium · avg 42%▇▇ 4.0%at its level
Arena-Hard v2.0 creative writing (GPT-4.1 + Gemini-2.5 judges)60.9% ✎▆▂▂▆▂▂▃█▂▃·▂ beats 16/22hard · avg 19%▇ 1.8%sharp · at its level · noisy
IFEval prompt-level loose85.8% ✎▁▃▁▃▁▄▁▁█▃▃▁ beats 19/22easy · avg 80%▇ 0.8%typical
math▁▂▃▅▆▅▄█▄▁▁▁ #146 of 398 · 9.3% of its score · 3 tests

Beats Claude Sonnet 4.6, MiMo-V2.5 (310B-A15B), MiniMax-M2.7 here, though they rank higher overall. Trails Nemotron 3 Nano 30B-A3B, Olmo 3 32B Think, Olmo 3.1 32B Think here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
OTIS mock AIME 2024–202559.2% ✎▃▃▄▃▃▃▇▇█▃▂▅ beats 89/193very hard · avg 8.5%▇▇▇ 6.4%sharp · at its level · noisy
MATH (subset unspecified)98.1% ✎▃▃█▂▂▆·····▃ beats 22/22medium · avg 61%▇ 1.7%typical
GSM8K95.5% ✎▂▃··▃▂▅█▆▃▂▃ beats 22/22very easy · avg 88%▇ 1.3%sharp · maxed out
other tests10 not in a task · 37% of its score
testresultvs the others in its rundifficultycountswhy
GPQA Diamond65.3% ✎▁▁▁▁▂▄▆▄▅▇█▅ beats 94/208hard · avg 40%▇▇▇▇▇ 12%sharp · at its level
MMLU88.4% ✎▁▂·▁·▁▁█·▃▃▅ beats 22/22easy · avg 77%▇▇ 4.9%typical
BIG-Bench Hard89.6% ✎▁▄·▁▁·▅█·▄▄▃ beats 22/22easy · avg 75%▇▇ 4.7%sharp
EQ-Bench 3 (emotional intelligence, judge Elo)1197.5 ✎▁·▂▁▄▅▅▃█▄▁▁ beats 30/62avg 1044▇▇ 3.9%typical
ARC-AGI-1 public evaluation set (ARC Prize model_baseline, pass@2)11.3% ✎▁·▇▃▁·█▇▁·▇▃ beats 3/27very hard · avg 4.1%▇ 3.7%sharp · noisy
BRIDGE clinical text, zero-shot average (87 tasks)39.4% ✎▁▁▁▁▁▂▂▃▄▆█▃ beats 67/97hard · avg 32%▇ 3.5%blunt · at its level
Open Persian LLM Leaderboard v2 average (21 tasks)60.6% ✎▂▄▂▃█▄▃▆▃▅▂▂ beats 52/81medium · avg 58%▇ 2.4%blunt · at its level
Chess Puzzles (Epoch)5.0% ✎█▁▁▁▁▂▁▄▂▂▁▁ beats 0/135very hard · avg 5.9%▇ 1.0%sharp · too hard for it · noisy
2 more tests
testresultvs the others in its rundifficultycountswhy
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy85.3% ✎▃·▁▁▂▁▂▅█▇▄▁ beats 84/100medium · avg 68%▇ 0.8%noisy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy93.2% ✎▁▁▁▁▁▁▁▂▄█▄▃ beats 129/146easy · avg 83%▇ 0.4%noisy

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run QwQ-32B · needs ~19.9 GB · from huggingface.co/Qwen (official)
    chat in the terminal
    llama cli -hf Qwen/QwQ-32B-GGUF:q4_k_m
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf Qwen/QwQ-32B-GGUF:q4_k_m

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run QwQ-32B · needs ~19.9 GB · from huggingface.co/Qwen (official)

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/Qwen/QwQ-32B-GGUF@q4_k_m -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run QwQ-32B · needs ~20.16 GB · from ollama.com
    chat in the terminal
    ollama run qwq:32b

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run QwQ-32B · needs ~18.4 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/QwQ-32B-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/QwQ-32B-4bit

Official weights and documentation: huggingface.co/Qwen ↗

$ how the score was worked out · Epoch AI — internal evaluatio… 19%, 2 OLMo 2 Furious (Team OLMo,… 17%, Aider polyglot benchmark runs 17% and 10 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
Epoch AI — internal evaluations3▇▇▇▇▇▇ 19%CC BY 4.0
2 OLMo 2 Furious (Team OLMo, Ai2) — authors' own evaluations of OLMo 2 Instruct and peer instruct models6▇▇▇▇▇ 17%CC BY 4.0
Aider polyglot benchmark runs1▇▇▇▇▇ 17%Apache-2.0
Arena-Hard-Auto v2.0-Preview leaderboard (repository README)2▇▇▇ 11%Apache-2.0
LMArena leaderboard dataset1▇▇ 6%CC BY 4.0
WebApp1K models leaderboard (onekq-ai)2▇▇ 6%MIT
CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models (arXiv:2503.16167v2)4▇ 5%CC-BY-4.0
EQ-Bench 3 canonical leaderboard results1▇ 4%MIT
ARC Prize Foundation model_baseline outputs on ARC-AGI-1/2 public evaluation sets1▇ 4%MIT
BRIDGE Medical Leaderboard (Harvard Medical School / Brigham and Women's YLab)1▇ 4%MIT
Berkeley Function Calling Leaderboard result archive2▇ 3%Apache-2.0
Open Persian LLM Leaderboard v2 (Part DP AI and AUT NLP Lab)1▇ 2%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇ 1%MIT
prior · pull toward the middle▇ 0%
weighted average= 59.4

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json